Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Take part in our daily and weekly newsletters to get the latest updates and exclusive content for reporting on industry -leading AI. Learn more
Hold me when you have heard this before: A little-known Chinese startup makes waves worldwide for an impressive new AI product.
No, we don’t talk about Deepseek R1, the KI argumentation model, that At the beginning of this year waves made between western circles. Instead, the hot new product is called you jour ManusAnd it is a new KI -Multipurpose agent -DH more than a KI model, it is an interface to control several models that can go out and even autonomously complicated tasks such as generating reports or executing dozens of social media accounts on behalf of the user.
If it sounds similar to the deep research modes that are offered by offers OpenaiPresent Googleand others as well as others Openais Operator Agent And Anthropic’s computer use mode (The latter two, like Manus, can take control of the computer or programs of a user, move the cursors and tap to carry out actions within software.
But for action-oriented leaders and decision-makers within companies in the West and abroad and abroad and chief technology officers, product managers, IT team leaders and a lot more-do you need to know about manus and the skills it offers? Read on to find out.
What is manus and who is behind it?
Manus Ai was officially Terminated on March 5, 2025 in the social network X With a contribution from the company that describes it as “the first general AI agent”, which autonomously explains complex tasks instead of just generating ideas.
After a Report on southern China Morgen Post (SCMP)Manus was developed by a Chinese company called Butterfly Effect who has offices in Beijing and Wuhan. According to reports, the company has only a few dozen employees, but quickly attracted attention in China’s AI landscape.
The founding team includes entrepreneurs and experienced product managers under the direction of Xiao Hong, a 33-year-old serial entrepreneur and graduate of the Huazhong University of Wuhan and Technology by Wuhan.
Xiao has previously created Wechat-based applications that have been taken over by larger companies and later launched Monica.ai, an AI assistant, which is available as a browser expansion and mobile app.
On his website, Manus should be named after the Latin word for “hand”, an allusion to the fact that users can rely on the fact that they can carry out actions for them, or in my words to “lend you a hand”.
How does Manus Ai work?
Manus AI is designed as a multi-agent system, which means that it combines several AI models to do tasks independently.
In contrast to AI chatbots that support users by providing information, manus can research, analyze data, generate reports, automate workflows and even write and provide code.
According to X-Posts by Ji Yichao, co-founder and chief scientist of Manus AI, the system is based on Anthropics Claude 3.5 Sonett -At this point in a nine-month-old AI model and finely coordinated versions of Alibabas Qwen models.
The team is currently testing an upgrade from Manus The latest and most powerful model from Anthropic, Claude 3.7, This is expected to further improve its argumentation and execution skills.
Manus Ai works asynchronous, which means that users can assign tasks and go away while they do them autonomously. It is currently in private beta, whereby access is granted by invitation codes.
How does Manus Ai stack against the competition based in the USA?
One of the main reasons why Manus Ai has gained traction is the strong benchmark performance. According to a graphic published on the official manus website, this was published on the official manus website, and the “former state-of-the-art” from Openas O3-Note-Deep research agent and the “previous state-of-the-art”.
This claim, together with real tests, has led some AI power users and early users to the conclusion that Manus could possibly be one of the most capable autonomous AI agents that are available today.
In addition to the benchmarks, Manus has already proven on freelance platforms such as Upwork and Fiverr and at Kaggle machine learning challenges in order to successfully carry out complex real tasks.
Ki -Influencer celebrate the arrival of Manus and the impressive performance
The conversation about manus in the media and circles started at the end of last week when User on social network X found that some people used it to automate the management of Up to 50 social accounts at the same timeIn real time to show the skills, to create fleets of engagements that could use companies for reviews.
Although this has not yet been proven for Manus, the same technology could probably be used for all types of marketing and influencing campaigns, even for political propaganda or disinformation.
But for the most part, Ki -Power users and influencers in the west were largely impressed and celebrated Manus’ arrival -impressed by their first tests as soon as they received some of the tight beta invitations or observed the work of others with access to the tool.
Rowan Cheung, founder of The dilapidated AI newsletterdescribed Manus Ais Start as a potential turning point for AI agents and said: “China’s second deepseek moment is here, in A contribution on his LinkedIn account.
“This AI agent called” Manus “is becoming crazy viral in China at the moment … it is like deep research + operator + claude computer together and it’s really good.”
Cheung personally tested manus and found that it:
- Created and prepared a biography website about itself with 100 percent accuracy and real-time data acceptance.
- First -class rental spots in San Francisco, which are based on crime rates, presence of the AI industry and the density of entrepreneurship.
- Developed a complete AI course and generated eight chapters with content, including tools, application cases and input requests.
He received 500 invitation codes from the Manus team and spotted them to his subscribers and readers.
Former Googler and AI aligned YouTuber Bilawal Sidhu announced a practical video evaluation and called Manus “the next thing I saw an autonomous AI agent.”
“It is like standing over the shoulder of a computer … You ask you what you should do at the highest level, and it basically does it for you.”
Sidhu tested Manus on various tasks, including:
- Research locations research – Scanning Google Maps and news sources to recommend the best places based on regulations, accessibility and security.
- Development of video applications-automatic video effects such as ASCII art and real-time filters.
- Extracting industrial knowledge – data from Reddit, Twitter and other sources to generate comprehensive reports.
Sidhu emphasized that Manus concludes complex workflows and step by step, what makes them one of the most autonomous AI agents that are available today.
Another AI influencer, Chubby (@kimmonism on x)claimed Manus Ai exceeded Openai’s deep research: “I am not paid for the promotion. I just tried the preview access and it exceeded Openai’s deep research by a lot! Stunning. “
Andrew Wilkinson, co-founder of the startup exit provider and purchaser Tiny, wrote on x This is absolutely crazy. I have the feeling that I am only six months into the future. I threw him a zip file of 20 applicants for a CEO job and it did a deep dive each … “
Great questions about Manus’ future remain
Despite the excitement, questions and concerns about the true skills of Manus Ai have arisen.
According to the South China Morning Post, very few people were able to test the Manus -KI due to the serious lack of server after the sudden increase in fame.
Zhang Tao, product partner of Manus AI, indicated in a social media contribution: “The current mechanism only invitation is due to a really limited server capacity at this stage. The team underestimated the enthusiasm of the public answer, and our server resources were only planned for a level of demonstration. ”
This lack of availability has fueled speculation that Manus Ai may use scarcity marketing tactics to create hype. Accordingly, invitation codes for Manus were sold to China’s second-hand marketplace Xianyu, which led to frustrations among potential users.
In addition, users like @taxestex to X (the self -described “Stan #1” or obsessive fan from Deepseek) said that “it is a product that has been optimized for influencers, which is why it exploded so much. Creation of threadboy content, trigger plans and such general interest (mind blown emoji, hand that emoji indicates) stuff – yah. Stem help, coding – worse than googling. More LLM than agent. “
In other words, some believe that it is good for influencers to build a followers through spam lists and posts, but not as well as other tools at work like development. Even a former Venturebeat writer and current Techcrunch-Er Kyle Wiggers Said in one piece over the weekend: “Manus is probably not China’s second” Deepseek moment “.”
Is Manus ai the real deal?
In contrast to Deepseek R1, which has developed its own foundation model, Manus Ai relies on existing large language models (LLMS) such as Claude and Qwen. This prompted some critics to ask how much of Manus’ technology is really original.
Yichao Ji recognized this in the official introductory video and explained: “Manus works as a multi-agent system that is powered by several different models.”
Nevertheless, Ji emphasized that Manus Ai is obliged to develop the open source and forwards some of his models to the public later this year.
At the moment, Manus Ai is only available in private beta via invitation codes. The company plans to expand access soon, but the restrictions of the servers can delay a complete public start. An open source component is planned that enables developers to integrate parts of their AI system.
While some Manus Ai consider a breakthrough in autonomous AI agents, others remain skeptical about their true skills, its business model and its scalability. While the hype around Manus Ai continues to grow, the question remains: Is this the next deepseek moment or just a checked AI product?
